中文摘要：

针对泵站优化运行计算时最优解评价指标单一的问题，建立了包含机组开停机约束的泵站优化运行数学模型和运行方案选优的投影寻踪决策模型。提出了求解泵站多机组优化运行模型的蚁群算法，并通过分析模型的特性改进了算法中启发式信息和信息素更新方式。对江都四站多机组Ft优化运行计算的结果显示，变量同等离散的情况下，利用蚁群算法优化的结果比用动态规划逐次逼近法优化的结果节省了2．8％的电费，前者相比设计工况运行时节省了29．2％的电费，且蚁群算法优化结果对应的运行方案中叶片调节次数少，机组运行时间短；方案选优时投影寻踪决策模型能够兼顾多个评价指标的优选，得到的运行方案不仅运行成本低，且更合理，更贴切于日常运行．可见改进后的蚁群算法结合投影寻踪决策模型在泵站优化运行及相近的领域有较大的实用价值。

英文摘要：

Usually the evaluation index of optimal pump operation solution is single. An optimal pump operation model which contained the constraint of start-stop pump unit was developed, and projection pursuit evaluation method for scheme optimization was proposed. The ant colony optimization algorithm was used to calculate the model. The heuristic information and the pheromone trail update method were improved by analyzing characters of the model for better performances. A calculation example for the No. 4 Jiangdu pumping station was conducted. The results from ant colony optimization algorithm showed that 29.2% of energy fee could be saved under the designed operation condition, which was compared with the result from dynamic programming with successive approximation algorithm under the same discrete condition, and was better with 2.8% of the result from dynamic programming. The results from ant colony optimization algorithm had less times of the blade adjusting, and shorter operating time of the pumps. Projection pursuit evaluation could give consideration to multiple evaluation indexes. The evaluation result of projection pursuit not only has low electric fee, but also suitable for practical operation, which indicates that ant colony optimization algorithm combined with projection pursuit evaluation method can provide a high application value to the field of optimal operation of pumping stations and related fields.